Agents could prove they act for a specific human

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

1. 5:38

   use cases what did you guys announce and what's that what's that supposed to solve yeah so there's the thing about ai is like it comes with some problems right we have to deal with them yeah but also comes with some really cool new things and one of those is agents right i think that people sometimes think that because we are like the company that is like all pro humans that we're against bots actually like agents are like awesome i use them all all the time every day and i i think that as the months and the weeks honestly continue to go by um whether it's muse dots whatever it is more and more of our usage of the internet of computers will be through agents and so the challenge companies are not having, including banks, is they are getting a tsunami of bots coming to their platforms. And it used to already be hard to know what was a human, what was a bot,

2. 6:26

   what's a bad bot, what's a good bot. Now it's way harder to tell. There's way more of those. So they're having to face really hard decisions. They either fully lock down their platforms. For example, JP Morgan Chase just announced, and I believe that even Citigroup or Bank of America yesterday announced that they're fully going to block out bots. So that's the safe way to go. But the problem is consumers now expect that we are able to use our agents. And what's going to happen is then all of us consumers, we're just going to go and start using other banks that let us do that. There was this very famously, this happened with Amazon a couple of weeks ago, where Paul Graham was tweeting, I think this is the first opening to compete with Amazon in this whole history, is because they're blocking agents. Someone will build a marketplace that allows

3. 7:08

   agents to transact there and now to do that you have to figure out how to do that safely and so what we think is the answer is uh being able to delegate your proof of human to your bot and so what that looks like is almost like giving your bot power of attorney where you're able to say okay my muse it's able to go on these websites do certain things and act on my behalf and so when it shows up when it shows up to amazon or wherever it is it shows up and it doesn't pretend to be you it very clearly says like hey i am a bot but i am representing a human and not just a random human i'm representing tiago right and and so if that mod misbehaves that website can say block all bots from tiago in the future right or it can give him a timeout or whatever or he can give you a special discount because tiago has never shopped on this store uh and so maybe i'll give you a 30 discount

4. 7:54

   but then if tomorrow you use chat gpt and you try to get the same discount you won't be able to do it and so it just opens up a whole world is this protocol agent kit of unlocking the potential of agents because it lets you make the most of good bots while preventing the risks of bad bots. I want to shout out our sponsor Lovable. You know the app you've been meaning to build, the internal tool, side project, or product